Centralised Smart EV Charging in PV-Powered Parking Lots: A Techno-Economic Analysis
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper proposed a centralised V1G smart charging (SC) algorithm for EV parking lots considering real EV charging dynamics which minimises both the EV charging costs for their owners and the CPO electricity provision costs, or the related CO 2emissions. An innovative SC benefit splitting algorithm that makes sure the SC savings are fairly split between the EV owners is also introduced. However, the paper has the following problems that need to be further modified:
1) For Table 1. Relevant literature overview on smart EV charging techniques, the table is relatively single for the main technical literature review, such as Minimising System Three MILP methods in Cost are listed only as references [16]~[18].
2) For Costs and Emissions, Eqs. (25) and (26), how to consider these final charging cost in optimization should be carefully explained.
3)How this smart charging optimization method is applied in practical engineering? Can the author add empirical research?
Author Response
Comment 1: For Table 1. Relevant literature overview on smart EV charging techniques, the table is relatively single for the main technical literature review, such as Minimising System Three MILP methods in Cost are listed only as references [16]~[18].
Response 1: In the updated version of the manuscript, seven additional references, namely [8-9], [13-15], and [29-30], were added and explained in the text following the literature review table. All these papers came out between 2023 and 2024, so we believe that they represent a valuable and updated contribution to the available literature on the topic. References [13-14], and [30] propose complex MILP problems, and supplement the already available ones.
Comment 2: For Costs and Emissions, Eqs. (25) and (26), how to consider these final charging costs in optimization should be carefully explained.
Response 2: The dynamic costs from equations (25) and (26) are used in equation (1), where the objective function is defined. To make it easier for the reader, we moved equations (25) and (26) to the methodology section, and are now labelled as (6-7), so they are presented right after the definition of the objective function. Lines 137-138 were also added, to further clarify the meaning of the included DSO and TSO cost terms.
Comment 3: How is this smart charging optimization method applied in practical engineering? Can the author add empirical research?
Response 3: Figure 2 was added to better explain how this algorithm will be implemented in practical engineering. Empirical research is undergoing as part of the ongoing FLOW project but, since the tests have not been finalized yet, presenting their results is not within the scope of this paper. This paper is instead aimed at benchmarking the performance of the proposed algorithm through the use of historical data coming from one of the e-mobility service providers involved in the project.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe aper addresses the charging of EVs in parking lots.
As far as described in the paper an independent operator of the charging stations is assumed. However, later this is coupled with a PV-system on a nearby building. This scenario is more like the one seen in commercial settings or local area grids. Or is the PV data only used as an example to show the yield I this area and the calculation assumes that charging stations and PV-system are owned and operated by the same entity? Please clarify.
If all this is part of the university network, then most probably the assumption on page 12 “we assume that the installation is similar to a domestic one” will not hold. I
It looks as if only the energy price applies to the CPO. Often charging station operators are seen as end consumers and from a given annual consumption (grid consumption) depending on local regulation , a recording power measurement is provided for and then also billing according to peak consumption and energy (depending on the ratio in the year, also in different tariffs). In this case: From an economic perspective peak shaving is very interesting. See e.g.:
Meiers, J.; Frey, G. A Case Study of the Use of Smart EV Charging for Peak Shaving in Local Area Grids. Energies 2024, 17, 47. https://doi.org/10.3390/en17010047
However, the electricity cost model is not clearly described in the paper. Footnote 6 on page 12 links to a website in Danish language. Please explain in the paper.
Apart from that, the customer will only receive a price per kWh, but this normally will depend on the charging capacity class. I don't know what is included in the calculation here - only the technology or also the grid connection (construction cost subsidy) or a corresponding power tariff.
The rationale behind the statement on page 3 and the later saving splitting process is not clear to me. If charging time is as assumed given by the normal office hours, or by the user via an app, the EVs are fully charged before leaving, where is the experienced delay? It makes no difference at what time the EV is charged as long as it is ready at the time when it is disconnected. Hence why should there be a benefit? Or are there EVs that are not fully charged? Please explain and give data on this.
Your study is based on a maximum of 12 connected EVs. How do the results scale? In sense of #of charging points and size of PV? How do the algorithms scale in the # of charging points and # of EVs (or better charging sessions)?
Line 130: c_t^spot: As far as I know, costs for the direct marketer are deducted from the spot market price if the CGO is not one himself
Page 10: You state that circulating EVs have nominal charging power of 11kW. On the other side your statistics show the peak at 3.7kW. Can you explain this?
Technical Points:
V1G is not explained, V2B neither
Acronym BSA is used on page 3 but not introduced until page 7.
Maybe you should provide a list of acronyms
Figure 6: The figure does not provide the information it should: take 16;00 there are three types of green. What does it mean? Which one is power stop, which is the connection?
Author Response
Comment 1: As far as described in the paper an independent operator of the charging stations is assumed. However, later this is coupled with a PV-system on a nearby building. This scenario is more like the one seen in commercial settings or local area grids. Or is the PV data only used as an example to show the yield I this area and the calculation assumes that charging stations and PV-system are owned and operated by the same entity? Please clarify.
Response 1: The reason why we include the PV system is that, by far, it is the greenest and most cost-effective way of reducing the charging cluster operational cost for the CPO. As the reviewer correctly understood, both the PV system and the EV charging stations are owned by the charging point operator (CPO), which is the university campus service. We made it clearer both in the abstract (lines 11-13) and case study (line 268) sections, so that our assumptions are easier to understand.
Comment 2: If all this is part of the university network, then most probably the assumption on page 12 “we assume that the installation is similar to a domestic one” will not hold.
Response 2: We thank the reviewer for this comment. Indeed, in the specific area where the parking lot is located, the grid tariffs should be of the “B-low” type, which includes customers connected to the LV grid, but without a dedicated transformer. However, given the low EV charging capacity installed in the parking lot, the cluster was considered as a “domestic” one, i.e. the “C” type. We repeated all the simulations for the B-low customer type, and the results do not change significantly. In any case, we made this clearer in the Case Study section, at line 268.
Comment 3: It looks as if only the energy price applies to the CPO. Often charging station operators are seen as end consumers and from a given annual consumption (grid consumption) depending on local regulation, a recording power measurement is provided for and then also billing according to peak consumption and energy (depending on the ratio in the year, also in different tariffs). In this case: From an economic perspective peak shaving is very interesting. See e.g.: Meiers, J.; Frey, G. A Case Study of the Use of Smart EV Charging for Peak Shaving in Local Area Grids. Energies 2024, 17, 47. https://doi.org/10.3390/en17010047. However, the electricity cost model is not clearly described in the paper. Footnote 6 on page 12 links to a website in Danish language. Please explain in the paper.
Response 3: Currently, in Denmark, only industrial consumers are paying a power-related cost, based on an average of their 10 peak consumption values throughout the previous 12 months. Following the assumptions explained for point 2), these tariffs were not applied, since the installation is considered as a domestic one. In addition, we agree that peak shaving is very interesting, and we also make sure that the power does not exceed the parking lot contracted capacity with constraint (10). The link in the footnote takes the reader to the page where the different tariffs are defined by the local DSO, Radius. In the footnote, we made it clear for the reader what the page shows, so the language barrier is removed. Finally, the cost model is only accounting for a volumetric tariff based on the amount of withdrawn energy, as described in equations (5) and (6).
Comment 4: Apart from that, the customer will only receive a price per kWh, but this normally will depend on the charging capacity class. I don't know what is included in the calculation here - only the technology or also the grid connection (construction cost subsidy) or a corresponding power tariff.
Response 4: We are not sure what “charging capacity class” means, in this context, but we will assume it is the amount of energy per year that the customer absorbs from the system. Currently, in Denmark, the choice of a “class”, from C to A+, depends on the type of connection to the power system (i.e., LV/MV/HV and with/without a dedicated substation). Since this is considered a domestic customer, there are no power-related tariffs. The construction costs, together with the related subsidies, are also not included in the calculation. This choice was made since evaluating the payback period of the EV charging stations purchase is out of the scope of this paper, which is aimed at benchmarking the smart charging algorithm performance. We made this clearer for the reader at lines 334-340.
Comment 5: The rationale behind the statement on page 3 and the latter saving splitting process is not clear to me. If charging time is as assumed given by the normal office hours, or by the user via an app, the EVs are fully charged before leaving, where is the experienced delay? It makes no difference at what time the EV is charged as long as it is ready at the time when it is disconnected. Hence why should there be a benefit? Or are there EVs that are not fully charged? Please explain and give data on this.
Response 5: Since we are operating under the “perfect foresight” assumption, there is no unserved energy. If the smart charging algorithm is applied though, the EVs are fully charged at a later time, compared to the “uncontrolled” charging case, where no smart charging is applied. Since an EV owner that requires a small amount of energy might have to stay for longer, since maybe another EV requires a lot of energy in a shorter timeframe, we think that part of the saving that the second EV realizes should be distributed to the first. Hence, the “benefit splitting” algorithm is a way to make sure that the customers that had the most delay gets the higher share of the economic saving produced by the smart charging algorithm. We acknowledge that the explanation is not clear enough in the paper, so we clarified with an example at lines 94-97.
Comment 6: Your study is based on a maximum of 12 connected EVs. How do the results scale? In sense of #of charging points and size of PV? How do the algorithms scale in the # of charging points and # of EVs (or better charging sessions)?
Response 6: We thank the reviewer for this comment, that sparked our curiosity to try and test what would happen in both cases. To this end, we performed additional simulations, and added two more figures, representing a sensitivity analysis with respect to the installed PV capacity, and the average number of daily charging sessions in the parking lot. The reason why we based our analysis on the assumption of S6 is that the scenario represents the best case for the CPO, the PV system owner, while slightly reducing the charging costs for the EV owners. These results are presented in section 5.3.1 at lines 448-475, and underlined in the conclusions as well (530-535).
Comment 7: Line 130: c_t^spot: As far as I know, costs for the direct marketer are deducted from the spot market price if the CGO is not one himself.
Response 7: We are not familiar with the terms “CGO” and “direct marketer”, but we will try to clarify the cost structure for this parking lot in the following. The CPO (Charging Point Operator) pays the electricity provider (who operates on the day-ahead market) a cost equal to the energy provision cost, which is the sum of the spot market price, the DSO and TSO tariffs, and the taxes. The retailer is then responsible for paying the DSO and TSO tariffs to the respective entities, while the taxes are governmental. In this case, we do not consider an extra cost for the energy retailer, since it would be extremely difficult to estimate it (it highly depends on the contracted energy provision product). Nevertheless, this is a conservative assumption, because a higher CPO cost would increase even more the profitability of an algorithm that uses the local PV production to charge up the EVs.
Comment 8: Page 10: You state that circulating EVs have nominal charging power of 11kW. On the other side your statistics show the peak at 3.7kW. Can you explain this?
Response 8: Even though the majority of modern EVs can charge in three-phase at 11-22 kW, the ones that are charging at these spots are older models that are limited to 3.7 kW. Since this is a workplace parking lot, the people that charge at the cluster are almost always the same throughout the year. Hence, the high percentage of 3.7 kW charging sessions. We clarified this aspect for the readers at lines 292-294.
Technical Points:
Comment 9: V1G is not explained, V2B either.
Response 9: V1G is now explained in the introduction, and refers to unidirectional smart charging, where the EV power is reduced from the nominal one to slow down the charging session during expensive time slots. V2G is now defined in the methodology section, at line 170. V2B was never used as an acronym.
Comment 10: Acronym BSA is used on page 3 but not introduced until page 7.
Response 10: The phrase has been removed, so the acronym is defined in page 7 first, and used only afterwards.
Comment 11: Maybe you should provide a list of acronyms.
Response 11: We thank the reviewer for the suggestion, but we believe the paper is already quite long as it stands, so introducing an acronyms list would further increase its length. However, we double checked the acronyms again, to make sure everything is in order, and they are all defined before being used.
Comment 12: Figure 6: The figure does not provide the information it should: take 16:00 there are three types of green. What does it mean? Which one is power stop, which is the connection?
Response 12: We added an explanation to the figure caption: “Different shades of the each colour represent the overlapping parts of the histograms”. Indeed, some of the histograms overlap, and inevitably create different shades of the same colour.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThank you for sucessfully adressing all reviewer comments in this revised version.
